Keywords: Long COVID; children and adolescents; primary care; narrative medicine; semantic phenotyping; Human Phenotype Ontology; functional impairment; COOP/WONCA; family medicine

Background:

Long COVID in children, adolescents, and young adults remains under-recognized in primary care, despite growing evidence of persistent and disabling symptoms following SARS-CoV-2 infection. Fatigue, post-exertional malaise, cognitive dysfunction, and marked functional decline are frequently reported, yet these manifestations are often poorly captured by standard diagnostic frameworks.

Research questions:

To present a case-based method for identifying and characterizing Long COVID in individuals aged $\leq$ 20 years at the time of first SARS-CoV-2 infection, integrating narrative medicine with standardized digital phenotyping using the Human Phenotype Ontology (HPO).

Method:

From a longitudinal primary care cohort of over 400 patients with suspected Long COVID followed in Belgium between 2021 and 2025, 35 patients aged $\leq$ 20 years at first COVID infection were identified. Longitudinal family medicine records were analysed using a multimodal approach combining narrative clinical interpretation, semantic phenotyping based on the Human Phenotype Ontology (HPO), COOP/WONCA functional health charts, and the clinician-rated DUSOI severity index. HPO codes were used as a structuring tool and were not independently verified.

Results:

Patients presented heterogeneous, multisystem symptom profiles with prominent neurological, cognitive, pain-related, autonomic, fatigue-related, and exertional manifestations. Semantic mapping enabled transformation of real-world clinical narratives into structured phenotypic profiles while preserving clinical context. COOP/WONCA assessments revealed severe to extreme functional impairment in most patients, notably affecting daily activities, school attendance, and social participation. No meaningful correlation was observed between clinician-rated disease severity and patient-reported functional impairment.

Conclusions:

A substantial delay between acute SARS-CoV-2 infection and first clinical pickup was observed, with some patients accessing care only in late adolescence or early adulthood. Long COVID occurring before the age of 20 appears as a complex, multisystem condition with major functional impact that may remain unrecognized for prolonged periods in primary care. Integrating narrative medicine with semantic phenotyping may improve recognition and characterization of Long COVID in younger populations.

Points for discussion:

Think Long COVID early: Persistent fatigue, cognitive or exertional symptoms after COVID in young patients should raise suspicion, even without clear biomarkers.

Focus on function: Severe impact on school, daily life, or social participation may exist despite low clinician-rated severity.

Use continuity and narrative: Longitudinal family knowledge and simple functional tools help identify delayed and complex presentations.